This project offer a python script that help you setup diffusers training environment and also do training quickly, support the following train methods:
[1] [GPU]dreambooth-lora
[2] [GPU]dreambooth
[3] [GPU]text-to-image
[4] [TPU]dreambooth
[5] [TPU]text-to-image
Note:
- Tested on Ubuntu and TPU v3-8, will not work on Windows
- The tpu training only work on Google TPU v3
Make sure you have python3 installed. Check that by:
python3 --version
Make sure you have questionary
and distro
to make this script function properly:
pip3 install questionary
pip3 install distro
Clone this repo:
git clone https://github.com/Eric-liucn/diffusers_train_tools.git
Run the script
cd diffusers_train_tools
python3 main.py
One line command:
python3 --version && pip3 install questionary distro && git clone https://github.com/Eric-liucn/diffusers_train_tools.git && cd diffusers_train_tools && python3 main.py
This script is still work in progress, don't use it now...